More advanced projects than MIDIbox?

[intellijel]

I totally support the midibox project and admire the efforts of all those involved in maintaining a mandate to provide low cost, easy to use, open source projects.

However, I keep encountering limitations in using the PIC as a microcontroller and in using/adapting a base set of boards in a modular fashion.

I am interested in working on more advanced projects with more powerful uC's possibly more expensive (maybe evene surface mounted components) and designing custom boards.

Is there a forum for people developing more advanced music technology prototypes?

thanks,

    intellijel

[cheater]

There's the synth-diy mailing list if you're into that sort of thing...

Also I remember googling around for DSP and finding a mailing list on that. Don't know the link right now, though, sorry.

[stryd_one]

You don't *have* to use the standard MBHP modules... Look at c0nsumer's work for one example, and there are others out there ;)

If you're that serious, why not port MIOS to a more powerful processor for a 32bit core or something?

[intellijel]

ha! I am already on the Synth-DIY list (excellent, really helpful community).

As for porting to a 32 bit uC, I think I would rather just port to a better family of uC's like the AVR Atmel chips.

My strengths are in hardware design and mechanical work, I am not so much of a programmer anymore (but trying to rebuild my skills).

I would like to find something like the MIOS that already exists for another project. However, maybe porting the MIOS would be a good project for rehabilitating my dormant programming abilities.

cheers,

    intellijel

[stryd_one]

Well, the AVR isn't really that much more capable than the 18F... Search the forums for TK's comments on the AVR. Basically, you would spend a lot of time for very little benefit, if any at all.

Unfortunately a very tantalising comment was eaten on Friday the 13th, something about TK's ideas for the next generation core...

I would like to find something like the MIOS that already exists for another project.

I don't think that's going to happen... There's some groundwork laid out in FPGA synthesis, but that's audio not midi...

[TK.]

As a nice programming experience, porting MIOS to other microcontrollers isn't a bad idea. The OS itself isn't tied that much on the uC architecture (so long the applications are programmed in C). I for myself would never spent the effort to write a port for just another 8bit micro, but intellijel: if you want to do this for the AVR, the final results (e.g. performance) would be interesting!

Best Regards, Thorsten.

[cheater]

intellijel: if you want to do this for the AVR, the final results (e.g. performance) would be interesting!

and more ram!

[stryd_one]

More RAM?

I mean, hey it's your time, but if I was going to rewrite MIOS for a new uC, I'd want to see a lot more benefit for my work than a light increase in any resource, be it RAM or processing power or IO....

Uhm what I'm saying is, why trade your Mercedes for a BMW? Trade it for a Ferarri! I think that analogy made more sense ;)

Edit: Although it would be interesting to see if the AVR was any faster!

How about the Propeller ? The multitasking capabilities could be REALLY useful for MIDI... Not sure how widely available it is though, and it's very new, so it might be a dead technology in the future... I'm not a good person to recommend a suitable IC but there was a lot of hype around this...

[mess]

What about porting to the dsPICs?

(I think there was a thread about this in the past)

seems like an interesting port if you use the dsp features :-)

also it's a 16 bit architecture so maybe it could give some extra performance...

the alternative could be a pic18f4620 for control surface

and a dsPIC for the signal stuff

[cheater]

More RAM?

Doesn't the MB have a history of maxing out the PIC memory already?

by "more RAM" I meant orders of magnitude, not something like 10%!

[DrBunsen]

How about the Propeller

Indeed.  There have been some promising developments

[cheater]

I think Propeller isn't sure enough to invest time into learning of it.

Think about some bigger, more common families.

Maybe something surface-mounted. Those shouldn't be hard to assemble provided you have a ready PCB... I've seen a video tutorial here on the forums recently where a guy soldered a 96 pin chip in less than a minute.

[moogah]

http://rubidium.dyndns.org/cgi-bin/mailman/listinfo/fpga-synth

I'd love to see some more people pic up an FPGA to deedle with.  I'm itching to dive into the one I've got when the 808's off the ground. 

I would generaly agree that porting MIOS to another uC would only be interesting..  I would not suspect it would be anything too exciting until we're dealing with a processor capable of generating polyphonic sound or processing audio at a decent rate.  Until then all that would be achieved is increasing the number of unused cycles in each application.  More RAM is always a good thing tho, but we've already got that with the latest PIC.

On a side note: TK, what are you using to burn your 4620's?  Has the PIC burner been tested to be compatible?  Should I do that?  I'm tempted to just buy a burner from microchip to bypass any headaches.. but that's not very DIY.. and who says there won't be headaches with commercial units?

Back on topic there are dozens of specialized chips like the propeller out there, and so far people seem to have a good track record of making them into interesting projects.  One thing that I've been thinking of is creating a MIOS driver for AD's crosspoint switches in preperation of a Midibox audio mixer ala 02R.  THAT would be cool ;)

[TK.]

TK, what are you using to burn your 4620's?  Has the PIC burner been tested to be compatible?

Yes, I'm using MBHP_BURNER for this, the 4620 and 4550 were the reason why I built this programmer, because P18 was the only software at this time which was reliable enough.

Best Regards, Thorsten.

[stryd_one]

I'd love to see some more people pic up an FPGA to deedle with.

Totally... But they're kinda expensive to get into :( Look at the price difference between a MBHP core and the FPGASynth. Ouch.

I would generaly agree that porting MIOS to another uC would only be interesting..  I would not suspect it would be anything too exciting until we're dealing with a processor capable of generating polyphonic sound or processing audio at a decent rate.

Totally disagree. This is MIDIBox :) I'm all for DIY digital synths (I'd love to render my K5K obsolete by my own hand, despite the sacrilege ;) ), and I know that audio manipulation is far more trendy than midi mangling in recent years, but MIDI still has it's place, and I think it would be a crying shame to lose the MIDI focus of a community like this :( Don;t get me wrong though - MIOS/MBHP is an awesome platform to be partnered up with an audio device, be it a SID or OPL or FPGA or...... AD switch matrix.......

More RAM is always a good thing tho, but we've already got that with the latest PIC.

Yeh not to mention the SRAM extension. The vX prototype has 4Mbits of SRAM. Yes, that's an M for Mega ;D All this talk of AVR chips with more RAM than a 4620... are there any? Is it a big difference or a few K?

One thing that I've been thinking of is creating a MIOS driver for AD's crosspoint switches in preperation of a Midibox audio mixer ala 02R.  THAT would be cool ;)

Totally agree :) You've been reading my wiki page haven't you ;) I think I posted about this a while back too... I've got the driver basically done (it's pretty simple, it's a great chip and programming them is a snap), but the board design for the audio part is the tricky thing... IIRC (it's in the datasheet I can't be bothered reading again heh) you need to have 3 layers at least, to avoid ground noise and crosstalk problems, and if you're using it for mixing you have to think about balanced I/O too, as well as it needing 3v not 5v... I actually think it might not be that hard for someone experienced in synth diy *stares at you, moogah* ;D but I don't yet have the confidence in board design to spend the money on a prototype I designed with my n00b skillz heheheh

Seen this page ?

I think this kind of module would be indispensable... as well as the mixer and FX routing, and automated audio patchbays, one thing that people never mention is automated analog modular synth patching and routing... Imagine a real analog version of a nord modular... and anyone with a modular synth can hook a midibox up to it for full recall and remote editing.... droool *falls down*

PS the JDM works with 4620s too

[moogah]

I think this kind of module would be indispensable... as well as the mixer and FX routing, and automated audio patchbays, one thing that people never mention is automated analog modular synth patching and routing... Imagine a real analog version of a nord modular... and anyone with a modular synth can hook a midibox up to it for full recall and remote editing.... droool *falls down*

I think it was the AnalogSystems modular patch system that used the AD75019 to do the job.  There are compliactions with using something like this for patch storage on a modular, as Dr. Moog said: it's the knob positions more than the routing that determine the sound, so you wold have to build your modules specificaly to have *everything* wired for CV.  Still tho, it would be a key peice in any modular system!  I can't remember the exact part number, but I think it's the AD8113 which is a *fully buffered* 16x16 switch, as opposed to the 75019, which would require external buffers to connect multiple sources to the same destination without tying them all together.  I've got a tube of the 75019's specificaly for creating a programmable, patchable synth based on the OB-1.  However, having some expirience with the 808 project I can say that finishing something like that is a looong term thing.  Starting with a simple MIDI controlled routing module seems a good place to start, and I've def. got the Eagle skills to do the layout and the equipment to run tests on a prototype for noise and crosstalk.  Sounds like a fun project for spring/summer this year  :)  The real fun comes in adapting a single module to play multiple roles for an input summer block, channel fader block and buss/aux routing block.

Hey!  Here is cool chip to research, firewire or USB controllers that can be used to transfer audio to and from a PC!  (I know, I know, it's not MIDI.. but it could do MIDI too!  :D)

PS the JDM works with 4620s too

... I hope your kidding >:( >:(

;)

[stryd_one]

... I hope your kidding

WinPIC800 Supports the MBHP JDM and will program 452's, 4620, 4550, 16F88, etc etc... And it's fast too :)

I'll PM you about this patching thing ;)

[Artesia]

MIOS on the standard core seems to be fine for most things (allbeit somewhat taxed). A very powerful system can be built using afew boards concentrating on dedicated tasks & synced in parallel ..take for instance the mbsid. as for doing advanced & intensive Digital Signal Processing ..may as well have chips which are designed to do this, dedicated to this task alone... as TK once pointed out to me sometime back. As for adding a midi front end for it, may as well just sling in a MIOS core to concentrate on that side, passing information back and forth.. without having to completely re-write entire firmwares to handle multi-tasking; doing MIDI & Heavily data intensive tasks at the same time.

However if MIOS was to be ported to another chip platform, may as well pick a product family which is vastly more powerful (with a choice, left on chips.. as to how much power) ..so that having access to this as an option for really heavy duty stuff does not leave you looking for even more power futher on.

Also parallel processing is probably -the- way to go, as midi / audio lends itself very well into being split into sepperate sub-modules worrying about their own jobs..

Which kinda wraps back & just justifys stacking cores togeather working in tandem, worrying about their own tasks for lack of hassle... and of course using specialised chips to deal with things like DSP when its called for.

[hmm.. part of this conversation reminds me of something i've been thinking through, but decided not to start discussing in full quite yet...]

Building something like a 'MIDIBOX Modular', a system which is entirely software configurable & routeable; its already pretty much feasable with the current cores.. just need to dedicate each one to a particular part of the overall task, save their data locally that they need to remember. and then have it all syncronise from one core dealing with the midi side of things.

An ammount of taskload a given core can handle for each task (midi frontend, Oscilator control / envs, Filters and system routing/mixing), could be assigned and defined.. then all you need to do, if you want to say for instance double up on the size of a filter bank... is to expand 'sideways' in an almost -INFINATELY- scalable fashon.. by linking it in series off the end of that dedicated task chain ..rather than go hunting for more power & memory ...and then end up dealing with running out of resources, and end up having to reach for & headbashingly code for, even more powerful chips -just- to keep it all in the one 'shoebox'.